Security is designed based on the system and network for which it applies to. There are many ways to monitor, alert, track, trace, lock and jail each system or network. I do not intend to cover them all here. But, I do intend to give a clear guideline, road-map and a starting point for whatever system and network you may administrate.

Customization is the first key. Being original prevents an intruder from coming up with work-arounds since to do this requires knowledge of the environment. Original (or custom) environments do not lend this kind of flexibility. When it comes to monitoring, the best policy is to "roll-your-own" tools. Commercial products are great for productivity and support and offer cost-effective solutions. However, they are known systems that intruders have access to as well, and the would-be intruder can find an exploitable configuration testing safely within their own network before touching yours.

Autonomy is the second key. Again, we are faced with cost-effective, ease-of-use; ease-of-administration, versus true security. Integrated systems are just that -- integrated. When designing any integration method (hardware or software), attention should be made to how one will affect the other if there is any type of accident or compromise. With autonomous designs, there are many more keystrokes to perform a task. We have been trying to pull away from manual labor and onto a world of automation and simplification so much that we are willing to sacrifice a simple but under-acknowledged concept for it. Humans are intelligent and applications are not. We can program an application to appear intelligent, but if you look at the source code it is all logical program. There is no emotion applicable to binary logic operation. Emotion is true randomness which is why we are having trouble creating a real random number and why security is not 100%.